Makes the necessary changes, including adding helper
funcs, to bring the CLI up to an integration-testable
state. The changes made in this commit should be
sufficient for most other CLI tests. Includes a single
test as verification.
* Add import order linter
Error messages aren't the most clear/helpful, but they will point you to
the lines that need modifying.
* Adjust max errors reported
* Fix lint errors
Selectors and scope information are used to create the correct populate methods. Populate methods changed to fit type abstractions definitions for iteration, querying, and collection population.
Kopia maintainer mentioned this was a better way to deal with it than
the old version. Mostly superficial changes to logic.
Always safe to flush the write session during an upload. That will
ensure write sessions are closed. Validity of the uploaded data will
be via reachability of the data in the repo.
Drawback is if there's a problem flushing the write session it will
return that error instead of the original error (if there was one)
In order to both maintain control of test env data, as
well as ensure open-source users are able to run the
integration tests on personal account data, the test
user must be derived from the test configuration, rather
than hardcoded within the tests.
The alias of testing to ctesting prevented auto-importing
of the testing library. This change (arbitrarily) renames
the package so that it doesn't collide with the core pkg
for "testing".
`Items()` starts populate method for collection for the `exchange.Collection` package. Test packages expanded to verify the functionality of this feature.
Cobra's RunE causes every error response to display the
CLI help menu for the current command, after printing the
original error itself. This made error outputs difficult to
read, and was generally unhelpful in most cases. The exit
func in Print now prints the error and closes the CLI without
showing the help menu.
In order for sdk users to build an operation using a repo,
without importing /inernal packages, the operation
options needs to be available via a /pkg package.
* return errors if selectors reduce all entries
During a `backup details` or `restore` command, if the
selectors produce 0 results, return an error rather than
running a no-op process.
* add jsonDebug, MinimumPrintable to interface
We want to optionally print print either debugging
(containing all internal data) or user-friendly json
outputs. This requires a new flag: json-debug, and
an extension to the Print interface: minimumPrintable.
Functions and interfaces moved to interact properly with connector package. Fields are no longer exported until the required functions are moved to the exchange package.
embedding a baseModel means that objects like backup
have a backup.StableID, but no backup.ID nor backup.
backupID. This could be confusing for users. This change
swaps the value and type names, so that baseModels
include an ID value of type StableID.
* Function and tests for returned unescaped elements
* Regression test and fix for empty segment bug
If a raw segment had no elements that had length > 0 or just didn't have
any elements it would still create a segment, throwing everything else
off. Explicitly test for that now.
* store backup operation results in the backup manifest
Adds backup operation metadata like the outcome
statistics and selector definitions to the backup manifest
entry. These additional details will appear when users
call `corso backup list`.
* reduce filter selector flags to single strings
Since it doesn't make sense for many of the filter-type
selector flags to receive multiple input values (ex: what
would --received-after date1,date2 result in?), the filter
flags are getting reduced to single-string values.
* add must-pass filters to selectors
Extends the selectors scope set to include Filters. This
allows users to define all-pass matchers (filters),
separate from any-pass matchers (includes), and global
exclusions.
When a caller constructs an Users() exchange scope,
instead of building a single scope that contains all of the
provided users and all of the data beneath, follow the
pattern of the other scopes and generate one scope per
user-category pair.
